A child and adolescent memory capacity evaluation system and method based on electroencephalogram and a medium

Abstract

The application discloses a kind of based on electroencephalogram of children and adolescents memory capacity evaluation system, method and medium, belong to electroencephalogram signal analysis and cognitive ability evaluation technical field.It is through the acquisition of the electroencephalogram signal and behavior data of subject in the process of executing working memory task, pre-processes electroencephalogram signal, extracts scalp electroencephalogram energy space-time distribution characteristics, brain network functional connection characteristics and cortical source activation characteristics corresponding to memory coding stage, memory storage stage and memory extraction stage, constructs full-dimensional space-time core features;Quantitative correlation model between electroencephalogram features and behavior index is established based on full-dimensional space-time core features and behavior data, and memory coding ability score, memory storage ability score, memory extraction ability score and memory capacity comprehensive score are obtained;Dynamic topographic map, dynamic brain network diagram and three-dimensional source activation map are generated simultaneously for visual display.The application provides technical support for objective evaluation of children and adolescents memory capacity.

Description

Technical Field

[0001] This invention belongs to the fields of electroencephalogram (EEG) signal processing, cognitive neuroscience and artificial intelligence assessment technology. It relates to an EEG-based system, method and medium for assessing children's and adolescents' memory abilities, and particularly to an EEG-based system, method and computer-readable storage medium for assessing children's and adolescents' memory abilities based on spatiotemporal feature analysis of EEG signals and brain-behavior association modeling. Background Technology

[0002] Childhood and adolescence are critical window periods for brain development and cognitive function shaping, exhibiting extremely high neuroplasticity. Memory ability, as a core component of the cognitive system, directly determines children and adolescents' learning efficiency, knowledge acquisition ability, and social adaptability, significantly impacting their academic performance, cognitive development, and lifelong learning potential.

[0003] Working memory is a core component of the memory system, responsible for the temporary storage, processing, and updating of information. It forms the basis for higher cognitive activities such as learning, reasoning, and problem-solving. The strength of working memory directly affects an individual's learning efficiency and cognitive performance, and its development level is closely related to children and adolescents' academic performance, attention control, and executive function.

[0004] The n-back task is a classic standardized experimental paradigm for assessing working memory capacity. Its core advantage lies in its ability to modulate memory load in a stepwise manner using a single parameter n, and the neurophysiological markers of each processing stage have been thoroughly validated by numerous studies. This paradigm induces the encoding, updating, retention, and retrieval processes of working memory by requiring subjects to judge whether the current stimulus matches the nth previous stimulus. Because the stimuli are in a continuous sequence, there is some temporal overlap in the above processing processes, which can generally be functionally divided according to the following time windows: Encoding and early update phase (0-300ms after stimulus presentation): Initial registration of sensory information and attentional screening are completed, corresponding to N1 / P1 evoked potentials in EEG, accompanied by an increase in power in the frontal lobe theta band (4-8Hz) (this increase is sustained throughout the entire working memory processing process, and the higher the workload, the more significant the enhancement).

[0005] Information maintenance and continuous updating phase (300ms after stimulus presentation to before the next stimulus): relies on the coordinated activation of the prefrontal-parietal-occipital working memory core network. In this phase, the power of the alpha band (8-12Hz) in the posterior occipital-parietal lobe decreases with increasing load to enhance the input of task-related sensory information; the power of the theta band in the frontal lobe continuously increases, coordinating information transmission and cognitive control between brain regions.

[0006] The retrieval and decision-making phase (300-600ms after the presentation of the probe stimulus): information retrieval and matching decision-making are completed. The hallmark EEG component is P300, whose amplitude reflects the amount of cognitive resources invested and whose latency reflects the speed of information retrieval and processing.

[0007] By comparing the load effects of 1-back and 2-back tasks, and combining synchronously collected behavioral indicators (accuracy, reaction time, false alarm rate, false alarm rate, and d-prime sensitivity index) with the above-mentioned multidimensional EEG characteristics, an individual's working memory capacity, resource allocation ability, and executive control level can be effectively assessed, providing a reliable dual benchmark indicator of behavioral and neurological indicators for multidimensional quantitative assessment of working memory capacity.

[0008] Dynamic assessment of working memory capabilities based on electroencephalography (EEG) in developing children and adolescents can promptly identify developmental abnormalities or delays in working memory, providing a scientific basis for early intervention and regulation. Strengthening the monitoring of working memory capabilities in children and adolescents and implementing targeted regulation based on assessment results is of significant value in ensuring their healthy cognitive development, improving learning efficiency, and reducing the burden on families and the education system.

[0009] Current assessments of children and adolescents' memory abilities mainly rely on traditional methods such as standardized paper-and-pencil tests, behavioral observation, and clinical scales. While these methods have a certain foundation, they also have significant limitations: they can only be evaluated based on external behavioral performance and subjective feedback, failing to reflect the neural mechanisms of brain memory encoding, storage, and retrieval; the assessment process is highly dependent on professionals, resulting in high costs and low efficiency, making it difficult to meet the needs of large-scale routine screening and dynamic tracking in schools; and the results are mostly qualitative grading, unable to achieve multi-dimensional, refined, and quantitative analysis, thus failing to support personalized educational guidance.

[0010] While existing EEG-based assessment methods can obtain objective neural indicators related to memory processing and compensate for the shortcomings of traditional behavioral assessments, they still have significant limitations: 1. Existing methods for assessing children's and adolescents' memory abilities mainly rely on behavioral performance for evaluation, which cannot objectively reflect the state of neural activity during memory encoding, information storage, and information retrieval.

[0011] 2. Existing assessment methods have a low degree of automation, and the implementation process relies on professional personnel to operate and interpret the results, resulting in high assessment costs and low efficiency, making it difficult to meet the needs of large-scale screening and long-term dynamic monitoring of children and adolescents.

[0012] 3. Most existing assessment results are presented in a qualitative grading manner, lacking the ability to conduct detailed quantitative analysis of each component dimension of memory ability, making it difficult to provide an effective basis for individualized cognitive training and educational intervention.

[0013] 4. Existing EEG assessment platforms mainly focus on general EEG signal analysis and lack the ability to analyze the spatiotemporal characteristics of EEG in children and adolescents' memory processing, making it difficult to comprehensively characterize the neural activity characteristics in the memory processing process.

[0014] 5. Existing EEG assessment systems lack a fully automated processing mechanism from data acquisition, signal processing, feature extraction, ability assessment to result display. They also lack intuitive and efficient visualization methods, which limits their practical application in educational assessment and grassroots screening scenarios.

[0015] Therefore, existing technologies struggle to comprehensively characterize the neural activity features involved in the encoding, storage, and retrieval of memories in children and adolescents. They lack effective correlation mechanisms between EEG features and behavioral indicators, and they are unable to achieve automated quantitative assessment and dynamic visualization of memory abilities. How to objectively characterize the neural activity features involved in memory processing in children and adolescents using EEG signals, and establish quantitative correlations between EEG features and behavioral indicators, thereby achieving automated, objective, and refined assessment of memory abilities, has become a pressing technical problem for those skilled in the art.

[0016] This invention can provide key technical support for research on children's cognitive development, early screening of learning difficulties, and personalized education guidance, and promote the transformation of memory ability assessment from subjective qualitative to objective quantitative, and from static assessment to multi-dimensional dynamic analysis. Summary of the Invention

[0017] The technical problem to be solved by this invention is that, in view of the lack of joint analysis capabilities of scalp EEG activity, brain network connectivity and cortical source activation state during memory processing in children and adolescents, the difficulty in establishing quantitative correlation between EEG characteristics and behavioral indicators, and the difficulty in achieving automated assessment and visualization of memory ability, this invention provides an EEG-based system, method and medium for assessing the memory ability of children and adolescents.

[0018] In view of this, the purpose of this invention is to provide an EEG-based system for assessing memory abilities in children and adolescents, comprising: EEG acquisition equipment is used to collect EEG signals from subjects during the performance of working memory tasks, and simultaneously collect behavioral data; The processing terminal is communicatively connected to the EEG acquisition device and is used to receive the EEG signals and behavioral data, and to perform memory ability assessment processing. A display device, connected to the processing terminal, is used to display the evaluation results; The processing terminal includes: The preprocessing module is used to filter, segment, correct baselines, remove artifacts, and configure reference electrodes on the EEG signal to obtain a clean EEG signal. The feature extraction module is used to extract the spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features corresponding to the memory encoding stage, memory storage stage, and memory retrieval stage from the clean EEG signal, and to construct full-dimensional spatiotemporal core features. The brain-behavior mapping module is used to correlate and fit the full-dimensional spatiotemporal core features with behavioral data to establish a quantitative correlation model between EEG features and memory encoding ability, memory storage ability and memory retrieval ability. The assessment module is used to assess the memory ability of the target subjects based on the quantitative correlation model, and obtain memory encoding ability score, memory storage ability score, memory retrieval ability score and comprehensive memory ability score. The spatiotemporal display module is used to generate dynamic topographic maps, dynamic brain network maps, and three-dimensional source activation maps based on the full-dimensional spatiotemporal core features, and to visualize them in conjunction with the memory ability assessment results.

[0019] Furthermore, the feature extraction module extracts full-dimensional spatiotemporal core features according to the memory encoding stage, memory storage stage, and memory retrieval stage respectively; The full-dimensional spatiotemporal core features corresponding to the memory encoding stage include the spatiotemporal distribution features of scalp EEG energy; The full-dimensional spatiotemporal core features corresponding to the memory storage stage include brain network functional connectivity features; The full-dimensional spatiotemporal core features corresponding to the memory retrieval stage include cortical source activation features.

[0020] Furthermore, the preprocessing module obtains the clean EEG signal through the following steps: filtering the original EEG signal using a 0.5Hz to 50Hz bandpass filter; dividing the continuous EEG signal into EEG data segments corresponding to preset time windows according to the stimulus event markers; performing baseline correction using a preset baseline time window; and identifying and removing artifacts such as electrooculogram (EOG), electrocardiogram (ECG), and electromyogram (EMG). Rereference processing is performed on the EEG signals.

[0021] Furthermore, the spatiotemporal distribution characteristics of scalp EEG energy include theta band power characteristics, alpha band power characteristics, and the amplitude and latency characteristics of event-related potentials; The brain network functional connectivity features are obtained through the functional connectivity matrix between brain electrical channels, and the clustering coefficient and global efficiency are calculated based on the functional connectivity matrix. The cortical source activation features are obtained through brain power imaging and include the cortical source activation time series corresponding to the target brain region.

[0022] Furthermore, the brain-behavior mapping module uses the ridge regression algorithm to establish the quantitative correlation model; The full-dimensional spatiotemporal core features are constructed into an EEG feature matrix; The behavioral data is then constructed into a behavioral target vector; By introducing an L2 regularization constraint term into the ridge regression objective function, the mapping relationship between EEG features and behavioral indicators is obtained, thereby reducing the impact of the correlation between the full-dimensional spatiotemporal core features on the model training results.

[0023] A method for assessing memory abilities in children and adolescents based on electroencephalography (EEG), comprising: Collect electroencephalogram (EEG) signals and behavioral data from subjects during the performance of working memory tasks; The EEG signals are preprocessed to obtain clean EEG signals; Extract the spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features corresponding to the memory encoding stage, memory storage stage, and memory retrieval stage from the clean EEG signals, and construct full-dimensional spatiotemporal core features. A quantitative correlation model between EEG features and behavioral indicators is established based on the full-dimensional spatiotemporal core features and the behavioral data. Based on the quantitative correlation model, the memory ability of the target subjects is assessed to obtain memory encoding ability score, memory storage ability score, memory retrieval ability score and comprehensive memory ability score; A visual display result is generated based on the full-dimensional spatiotemporal core features and the memory ability assessment results.

[0024] Furthermore, the extraction of cortical source activation features includes: Establish a brain power source imaging model between scalp EEG observation signals and cortical source signals; The observation residuals in the brain power imaging model are modeled as a Student t-distribution to characterize non-Gaussian noise and outliers in the EEG signal; The student t-distribution is approximated by a Gaussian mixture distribution, and variational parameters are introduced to construct the observation noise covariance matrix; Based on the observed noise covariance matrix and the prior distribution of the cortical source signal, the posterior distribution of the cortical source signal is obtained by Bayesian inference. The cortical source activation time series is determined based on the posterior distribution, and the cortical source activation time series is used as the cortical source activation feature.

[0025] Furthermore, the approximate representation of the student t-distribution using a Gaussian mixture distribution includes: Introduce variational parameters corresponding to each observation channel; The student t-distribution can be expressed as a multiplicative distribution of the Gaussian and inverse gamma distributions using the variational parameters. A diagonal observation noise covariance matrix is ​​constructed based on the variational parameters to characterize the noise level corresponding to different EEG channels, wherein the observation noise covariance matrix is ​​a diagonal matrix composed of the variational parameters of each observation channel.

[0026] Furthermore, the method of obtaining the posterior distribution of cortical source signals using Bayesian inference includes: Construct a probabilistic model with a zero-mean Gaussian distribution as the prior distribution of cortical source signals; The posterior mean and posterior covariance matrix of the cortical source signal are calculated based on the observed EEG signal, lead field matrix, observed noise covariance matrix, and the prior distribution. The model parameters are iteratively updated by maximizing the log-likelihood function of the complete data until the preset convergence condition is met. The converged posterior mean is used as the robust estimation result of the cortical source signal, and the corresponding cortical source activation time series is generated. The cortical source activation time series, together with the spatiotemporal distribution characteristics of scalp EEG energy and brain network functional connectivity characteristics, constitute the full-dimensional spatiotemporal core features, and are used to establish a quantitative correlation model between EEG features and behavioral indicators.

[0027] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the electroencephalogram-based method for assessing memory abilities in children and adolescents.

[0028] The beneficial effects of this invention are as follows: (1) By combining the spatiotemporal distribution characteristics of scalp electroencephalogram energy, brain network functional connectivity characteristics and cortical source activation characteristics, this invention achieves multidimensional characterization of the neural activity state of the entire process of memory encoding, memory storage and memory retrieval in children and adolescents, thereby improving the analytical ability of memory processing and the completeness of evaluation information.

[0029] (2) By establishing a quantitative correlation model between EEG characteristics and behavioral indicators, this invention realizes an objective mapping between neural activity characteristics and memory performance, and can obtain memory encoding ability score, memory storage ability score, memory retrieval ability score and comprehensive score, thereby improving the objectivity and quantification of memory ability assessment results.

[0030] (3) The present invention adopts a robust brain power estimation method based on the student t-distribution, Gaussian mixture distribution approximation and Bayesian inference, which can reduce the influence of electrooculography artifacts, electromyography interference and other non-Gaussian noise on the cortical source localization results, improve the stability and reliability of cortical source activation feature extraction, and is especially suitable for the EEG data analysis scenario of children and adolescents.

[0031] (4) This invention realizes an automated processing flow for EEG signal acquisition, signal preprocessing, feature extraction, brain-behavior association analysis, memory ability assessment and result display, which reduces the dependence on the experience of professionals, improves assessment efficiency, and is conducive to promotion and application in schools, educational institutions and grassroots screening scenarios.

[0032] (5) This invention visualizes the spatiotemporal evolution of brainwaves through dynamic topographic maps, dynamic brain network maps and three-dimensional source activation maps, which can intuitively present the brain activity change patterns of children and adolescents during memory processing, and generate a visualization report in combination with the assessment results, thereby improving the interpretability and display effect of the assessment results.

[0033] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0034] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the overall framework of the memory ability assessment system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of data processing in the memory ability assessment system according to an embodiment of the present invention; Figure 3 This is a flowchart of the memory ability assessment method according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the main interface of the children and adolescents' memory ability assessment system according to an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the effect of the spatiotemporal display module in an embodiment of the present invention; Figure 6 This is a schematic diagram of the process for assessing children's and adolescents' memory abilities according to an embodiment of the present invention; Figure 7 This is the memory task stimulation interface in an embodiment of the present invention; Figure 8This is an example diagram of a memory ability assessment report according to an embodiment of the present invention. Detailed Implementation

[0035] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0036] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0037] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0038] Please see Figure 1 This is a schematic diagram of the overall framework of the memory ability assessment system according to an embodiment of the present invention; Please see Figure 2 This is a schematic diagram of data processing in the memory ability assessment system according to an embodiment of the present invention.

[0039] 1. System Overall Architecture This invention provides a brainwave-based memory ability assessment system for children and adolescents, including an EEG acquisition device 100, a processing terminal 200, and a display device 300.

[0040] The EEG acquisition device 100 includes a behavioral data acquisition module 110, used to acquire raw EEG signals during the performance of a memory task by the subject according to a preset working memory task paradigm, and simultaneously acquire corresponding behavioral data. The EEG acquisition device 100 also includes a multi-channel EEG cap 120 conforming to the international 10-20 electrode deployment standard, such as a 32-channel, 64-channel, or 128-channel EEG acquisition device, to cover brain regions related to working memory, such as the prefrontal cortex, parietal lobe, and occipital lobe; the behavioral data includes indicators such as reaction time, accuracy, false negative rate, false positive rate, and the D-Prime sensitivity index.

[0041] The processing terminal 200 is communicatively connected to the EEG acquisition device 100, and is used to receive raw EEG signals and behavioral data uploaded by the EEG acquisition device 100, and to complete data preprocessing, feature extraction, brain-behavior association modeling, and memory ability assessment. The processing terminal 200 can be a desktop computer, laptop computer, workstation, server, or other electronic device with data processing capabilities, and internally runs a preprocessing module 210, a feature extraction module 220, a brain-behavior mapping module 230, an assessment module 240, and a spatiotemporal display module 250.

[0042] The display device is a high-resolution monitor used to present the stimulus interface, real-time feedback, and evaluation reports.

[0043] The display device 300 is connected to the processing terminal 200 and includes a task stimulus presentation module 310 for displaying the working memory task stimulus interface to the subject, and a result visualization module 320 for visually displaying the spatiotemporal dynamic features of EEG, the quantitative scoring results of memory ability, and the grade assessment results output by the processing terminal 200. The display device 300 may be a liquid crystal display, an all-in-one display terminal, a tablet device, or other terminal devices with graphics display capabilities.

[0044] In this embodiment, the EEG acquisition device 100 first acquires the EEG signals and behavioral data of the subject during the execution of the working memory task, and sends the acquired data to the processing terminal 200. The processing terminal 200 sequentially completes the EEG signal preprocessing, memory processing feature extraction, brain-behavior association analysis, and memory ability assessment processing to generate corresponding assessment results. Subsequently, the display device 300 displays the spatiotemporal dynamic evolution process of EEG and the memory ability assessment results in real time or offline, thereby realizing the automated assessment and visualization analysis of children and adolescents' memory abilities.

[0045] Please participate Figure 3 This is a flowchart of the memory ability assessment method according to an embodiment of the present invention; 2. Specific implementation of the preprocessing module The preprocessing module 210 is used to preprocess the raw EEG signals acquired by the EEG acquisition device 100 to obtain clean EEG signals for subsequent feature extraction.

[0046] Specifically, the preprocessing module 210 includes a signal filtering unit, a segmentation and baseline correction unit, an artifact processing unit, and a rereference unit.

[0047] The signal filtering unit is used to perform bandpass filtering on the raw EEG signal. Preferably, the signal filtering unit uses a bandpass filter of 0.5Hz to 50Hz to filter out DC drift, low-frequency baseline fluctuations, and high-frequency electromyography interference.

[0048] The segmentation and baseline correction unit is used to segment continuous EEG signals according to stimulus event markers in a working memory task. Preferably, the continuous EEG signal is segmented into EEG data segments from -200ms to 1000ms, with the stimulus presentation time as the zero point. Baseline correction is performed on each EEG data segment using a baseline time window from -200ms to 0ms to reduce the impact of baseline drift on subsequent analysis.

[0049] The artifact processing unit is used to identify and remove artifact components from EEG data. Specifically, the artifact processing unit can use independent component analysis to separate artifact components such as electrooculography (EOG), electrocardiography (ECG), and electromyography (EMG), and combine amplitude thresholding, gradient thresholding, or statistical anomaly detection methods to mark or remove abnormal EEG segments, thereby reducing the impact of non-brain-derived signals on memory ability assessment results.

[0050] The rereference unit is used to configure reference electrodes for the artifact-processed EEG signals. The reference electrode configuration methods include average reference, bipolar reference, or designated reference electrode rereference, to improve the signal-to-noise ratio and spatial comparability of the EEG signals.

[0051] Through the above processing, the preprocessing module 210 converts the original EEG signal into a clean EEG signal that is time-aligned, has reduced noise, and is suitable for feature extraction, and outputs the clean EEG signal to the feature extraction module 220.

[0052] The above processing can be implemented using Python's MNE-Python library, and users can also customize filter parameters and ICA components in a graphical interface.

[0053] 3. Specific implementation of the feature extraction module The feature extraction module 220 is used to extract full-dimensional spatiotemporal core features related to memory processing from the clean EEG signal output by the preprocessing module 210. The full-dimensional spatiotemporal core features include spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features.

[0054] Specifically, the feature extraction module 220 includes a scalp EEG energy spatiotemporal topology analysis unit, a brain network analysis unit, and a brain power imaging analysis unit.

[0055] The scalp EEG energy spatiotemporal topology analysis unit is used to perform time-domain, frequency-domain, and spatiotemporal topology analysis on clean EEG signals to obtain the spatiotemporal distribution characteristics of scalp EEG energy. In a preferred embodiment, the scalp EEG energy spatiotemporal topology analysis unit calculates the time-frequency power of different frequency bands using short-time Fourier transform, wavelet transform, or multi-window spectrum estimation methods. These frequency bands include theta band and alpha band; the theta band is from 4Hz to 8Hz, and the alpha band is from 8Hz to 12Hz. Simultaneously, the scalp EEG energy spatiotemporal topology analysis unit also extracts the latency and amplitude characteristics of event-related potentials (ERPs), and generates a dynamic topographic map of scalp energy based on the electrode positions of each EEG channel in the scalp space; it also extracts the latency and amplitude of ERPs (such as P300).

[0056] The brain network analysis unit is used to extract functional connectivity features of the brain network based on the phase synchronization relationship or statistical dependency relationship between multiple EEG channels. In a preferred embodiment, the brain network analysis unit calculates the phase lock value (PLV) between different EEG channels, constructs a functional connectivity matrix, and calculates network topology indicators based on the functional connectivity matrix. The network topology indicators include one or more of the following: clustering coefficient, global efficiency, node degree, and feature path length.

[0057] The brain power imaging analysis unit is used to invert scalp EEG signals to the cortical space to obtain the spatiotemporal evolution characteristics of cortical brain regions. In a preferred embodiment, the brain power imaging analysis unit performs source imaging processing on scalp EEG signals based on a preset head model and lead field matrix to obtain cortical source activation time sequences of the prefrontal, parietal, occipital, or other brain regions of interest. The brain power imaging processing can employ minimum norm estimation, the sLORETA algorithm, or the robust brain power estimation method further provided in this embodiment.

[0058] In this embodiment, the feature extraction module 220 calculates the aforementioned features according to different processing stages of the memory task. These different processing stages include the memory encoding stage, the information storage stage, and the information retrieval stage. Thus, the feature extraction module 220 can obtain the spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features corresponding to each processing stage, and combine these features to form a full-dimensional spatiotemporal core feature vector, which is then output to the brain-behavior mapping module 230.

[0059] 3.1 Robust Brain Power Estimation Method As a preferred implementation of the brain power source imaging analysis unit, this embodiment employs a source imaging algorithm based on the Student's t-distribution to improve the robustness of brain power source estimation to non-stationary noise and outliers, and to avoid the problem of source localization being susceptible to large-amplitude artifacts under the traditional Gaussian assumption. Compared to the Gaussian distribution, the Student's t-distribution has a thicker tail, is less sensitive to outliers, and can better fit the sudden disturbances present in actual EEG data.

[0060] Specifically, the probability density function of the model residuals Modeled as a student t-distribution: (1)

[0061] in, Let N be the observation noise at time t, N be the number of channels, and n be the channel ID. Let n be the observation noise at time t in channel n. The degree of freedom parameter affecting the tail thickness of the student t-distribution is... For gamma function, To measure the accuracy parameters of the noise.

[0062] At this time and The joint probability distribution no longer follows a Gaussian distribution. The observed signal at time t, Since the brain power source is at time t, this embodiment will follow previous research and approximate the Student's t-distribution according to the Gaussian mixture distribution (GMD). Therefore, this embodiment introduces a variational parameter in equation (1). Construct an approximate student t-distribution and rewrite it in the following form: (2) (3) (4) in, It is a Gaussian distribution. It is the conjugate prior of the Gaussian distribution variance, i.e., the inverse-Gamma distribution. It is the variational parameter of the GMD approximation of the Student's t-distribution, and it is independent of observation noise. .

[0063] Therefore, observation noise The variance can be rewritten as Assume the model coefficients follow independent and identically Gaussian distributions, where the mean is zero and the covariance matrix is... Therefore, the prior distribution of the model coefficients takes the following form: (5) Where M represents the number of brain power sources, and m represents the brain power source ID. Let be the prior variance of the m-th brain power source at time t. Let be the energy of the m-th brain power source at time t.

[0064] Then, in this embodiment, the model coefficients can be calculated based on Bayesian theory. The posterior distribution.

[0065] (6) in (7) (8) in, The EEG at time t for Model coefficients The posterior distribution of follows a Gaussian distribution with mean m and covariance matrix P.

[0066] For ease of writing, this embodiment defines all parameters in a parameter vector. In, that is (9) in, Let be the prior variance of the observation noise of the nth channel at time t.

[0067] Based on the above results, this embodiment can define the log-likelihood function of complete data as follows: (10) Wherein, parameter vector All model parameters can be obtained by maximizing the log-likelihood function of the complete data.

[0068] Through the above iterative updates, a robustly estimated cortical source activation time series can be obtained. Compared with traditional minimum norm estimation or sLORETA, this method has stronger resistance to outliers under non-Gaussian noise interference such as electrooculography and electromyography, and is particularly suitable for motion artifact scenarios commonly found in EEG data of children and adolescents. The cortical source activation time series output by this optimized method is used as the 'cortical source activation feature', which, together with the spatiotemporal distribution features of scalp energy and the functional connectivity features of brain networks, constitutes the core spatiotemporal features of all dimensions.

[0069] 4. Specific implementation of the brain-behavior mapping module The brain-behavior mapping module 230 is used to establish a quantitative correlation between EEG characteristics and behavioral indicators in order to achieve an objective representation of the subject's memory ability.

[0070] Specifically, the brain-behavior mapping module 230 receives the full-dimensional spatiotemporal core features output by the feature extraction module 220 and the synchronously collected behavioral data as input. The full-dimensional spatiotemporal core features include spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features; the behavioral data includes one or more of reaction time, accuracy, false negative rate, false positive rate, and the d-prime sensitivity index.

[0071] The brain-behavior mapping module 230 constructs the full-dimensional spatiotemporal core features into a feature matrix X, constructs behavioral data into a behavioral target vector y, and uses the Ridge Regression algorithm to establish a quantitative correlation model between EEG features and behavioral indicators.

[0072] In this embodiment, the feature matrix X, formed by fusing scalp energy, brain network indices, and cortical source features, is first preprocessed by normalization. Let y be the synchronously acquired working memory behavioral target vector (e.g., accuracy or reaction time of an n-back task). This invention establishes the following mapping relationship through ridge regression: (11) Where W is the weight coefficient matrix and b is the bias term. This is a regularization constraint term, which is introduced by the ridge parameter. This limits the excessive weighting of coefficients, thereby enabling the model to assign more stable parameter solutions when faced with highly overlapping feature information.

[0073] To obtain deterministic model parameters, the analytical formula for calculating the weight coefficients is as follows: (12) In the formula, I is the identity matrix. This is achieved through the characteristic covariance matrix. Add a small perturbation term to it. It can effectively handle the near-singularity problem of matrices caused by high correlation of features, ensuring the convergence and reliability of numerical operations.

[0074] Please see Figure 4 This is a schematic diagram of the main interface of the children and adolescents memory ability assessment system according to an embodiment of the present invention; 5. Specific implementation of the evaluation module The assessment module 240 is used to quantitatively assess the memory ability of the target subject based on the quantitative correlation model parameters output by the brain-behavior mapping module 230, and output the assessment results.

[0075] Specifically, the evaluation module 240 receives the full-dimensional spatiotemporal core features output by the feature extraction module 220 and the quantitative correlation model parameters output by the brain-behavior mapping module 230 as inputs. The full-dimensional spatiotemporal core features include spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features.

[0076] The assessment module 240 constructs a memory ability assessment model based on the parameters of the quantitative correlation model, and uses the memory ability assessment model to calculate and process the full-dimensional spatiotemporal core characteristics of the target subject to obtain assessment results related to memory ability.

[0077] In this embodiment, the evaluation results include at least: (1) Memory encoding ability assessment; (2) Memory storage capacity score; (3) Memory retrieval ability score; (4) Comprehensive score of memory ability.

[0078] The comprehensive memory ability score is calculated based on the memory encoding ability score, memory storage ability score, and memory retrieval ability score.

[0079] As one implementation method, when the quantitative correlation model adopts a regression model, the evaluation module 240 uses the parameters of the quantitative correlation model to predict and calculate the full-dimensional spatiotemporal core features of the target subject, and obtains the corresponding memory encoding ability score, memory storage ability score and memory retrieval ability score.

[0080] Furthermore, the assessment module 240 can perform level classification processing on the comprehensive score according to the pre-established level classification rules, and output the corresponding memory ability level result.

[0081] Preferably, the memory ability level result includes one of the following: excellent; good; average; needs improvement.

[0082] In one specific embodiment, the evaluation module 240 performs a comprehensive analysis based on the spatiotemporal distribution characteristics of scalp EEG energy in the memory encoding stage, the brain network functional connectivity characteristics in the memory storage stage, and the cortical source activation characteristics in the memory retrieval stage of the target subject. It outputs a memory encoding ability score of 85, a memory storage ability score of 72, a memory retrieval ability score of 88, and a comprehensive score of 81, and determines the corresponding memory ability level as "good".

[0083] The assessment module 240 sends the sub-scores, comprehensive scores, and grade results to the spatiotemporal display module 250 and the display device 300 for visual display, thereby realizing the automated assessment of children's and adolescents' memory abilities.

[0084] Please see Figure 5 This is a schematic diagram of the spatiotemporal display module effect in an embodiment of the present invention; 6. Specific implementation of the spatiotemporal display module The spatiotemporal display module 250 is used to visualize the full-dimensional spatiotemporal core features extracted by the feature extraction module 220 and the evaluation results output by the evaluation module 240, thereby intuitively presenting the spatiotemporal dynamic changes of brain neural activity during the memory processing of children and adolescents.

[0085] Specifically, the spatiotemporal display module 250 includes a dynamic topographic map display unit, a dynamic brain network display unit, and a three-dimensional source activation display unit.

[0086] The dynamic topographic map display unit is used to generate a dynamic topographic map based on the spatiotemporal distribution characteristics of scalp EEG energy. Specifically, it generates a scalp topographic map corresponding to each time point based on the EEG potential value or frequency band power value of each EEG channel at different time points, and displays them continuously in chronological order, thereby dynamically presenting the spatiotemporal evolution of scalp EEG activity. Preferably, the dynamic topographic map is displayed frame by frame in an animated manner, and the time step can be set according to actual needs, such as 10ms / frame.

[0087] The dynamic brain network display unit generates a brain network map based on the functional connectivity characteristics of the brain network. Specifically, it constructs a connectivity graph between brain regions based on a functional connectivity matrix, where nodes represent corresponding brain regions, node size represents the activation intensity of the brain region, connecting edges represent the functional connectivity between brain regions, and the thickness of the connecting edges represents the strength of the connectivity. Over time, the dynamic brain network display unit continuously updates the brain network structure to demonstrate the dynamic changes in the functional connectivity of brain regions during memory processing. Preferably, the brain network map is displayed using a force-directed graph layout.

[0088] The 3D source activation display unit is used to generate a 3D source activation map based on cortical source activation characteristics. Specifically, based on the cortical source activation time series output by the brain power imaging analysis unit, the activation intensity corresponding to different brain regions is mapped to the corresponding positions in a standard brain template, and a heatmap is used to display the activation degree of different brain regions, thereby dynamically presenting the spatiotemporal evolution of cortical brain region activation intensity. Preferably, the 3D source activation map supports rotation and zoom operations so that users can view the brain region activation status from different observation angles.

[0089] In this embodiment, the spatiotemporal display module 250 generates a dynamic topographic map, a dynamic brain network map, and a three-dimensional source activation map, and displays the above visualization results simultaneously through the display device 300, thereby demonstrating the spatiotemporal dynamic changes in scalp EEG activity, brain network connectivity, and cortical source activation state corresponding to the memory encoding stage, storage stage, and retrieval stage.

[0090] Furthermore, the spatiotemporal display module 250 is also used to display the memory encoding ability score, memory storage ability score, memory retrieval ability score, and overall memory ability level results output by the evaluation module 240, so as to realize the visualization display of the memory ability evaluation results.

[0091] Please see Figure 6 This is a schematic diagram of the process for assessing children's and adolescents' memory abilities according to an embodiment of the present invention; Please see Figure 7 This is the memory task stimulation interface in the embodiment of the present invention; 7. Example of evaluation methodology and process The following combination Figure 2 The present invention describes the method for assessing the memory ability of children and adolescents based on electroencephalography (EEG).

[0092] S1: Acquiring EEG signals and behavioral data Subjects wore multichannel EEG caps conforming to the international 10-20 standard and completed a preset working memory task on a display device 300. As a preferred embodiment, the working memory task employed an n-back task, including 1-back and 2-back tasks.

[0093] During the task execution, the EEG acquisition device 100 acquires the subject's EEG signals in real time and simultaneously records behavioral data. The behavioral data includes one or more of the following: accuracy rate, reaction time, false negative rate, false positive rate, and d-prime sensitivity index.

[0094] S2: EEG signal preprocessing The preprocessing module 210 receives the raw EEG signal and performs band filtering, segmentation, baseline correction, artifact identification and removal, and rereference processing on the raw EEG signal to obtain a clean EEG signal.

[0095] Specifically, the continuous EEG data is segmented into corresponding EEG data segments using the stimulus presentation time as a time reference point, and then output to the feature extraction module 220.

[0096] S3: Extract full-dimensional spatiotemporal core features The feature extraction module 220 analyzes and processes the preprocessed clean EEG signal to extract full-dimensional spatiotemporal core features.

[0097] The full-dimensional spatiotemporal core features include: (1) Spatiotemporal distribution characteristics of scalp EEG energy; (2) Functional connectivity characteristics of brain networks; (3) Cortical source activation characteristics.

[0098] Specifically, the feature extraction module 220 calculates the scalp EEG energy features corresponding to the memory encoding stage, the brain network connectivity features corresponding to the memory storage stage, and the cortical source activation features corresponding to the memory retrieval stage, and forms a full-dimensional spatiotemporal core feature vector to characterize the memory processing process.

[0099] S4: Establish a quantitative correlation between brain and behavior The brain-behavior mapping module 230 receives full-dimensional spatiotemporal core features and synchronously collected behavioral data, and uses a pre-trained quantitative correlation model to establish the correlation between EEG features and behavioral indicators.

[0100] Specifically, the brain-behavior mapping module 230 obtains quantitative correlation parameters corresponding to memory encoding ability, memory storage ability and memory retrieval ability based on the mapping relationship between EEG characteristics and behavioral indicators, and outputs them to the evaluation module 240.

[0101] S5: Generate memory ability assessment results The assessment module 240 performs assessments based on quantitative correlation parameters and the full-dimensional spatiotemporal core characteristics of the target subject, and outputs the memory ability assessment results.

[0102] The memory ability assessment results include: Memory encoding ability score; Memory storage capacity score; Memory retrieval ability score; Overall memory ability score; Memory ability level results.

[0103] The memory ability level result may include one of the following: excellent, good, average, or needs improvement.

[0104] S6: Visualize the evaluation results The spatiotemporal display module 250 generates a visual display result based on the full-dimensional spatiotemporal core features output by the feature extraction module 220 and the evaluation results output by the evaluation module 240.

[0105] Specifically, the spatiotemporal display module 250 generates dynamic topographic maps, dynamic brain network maps, and three-dimensional source activation maps, and displays the spatiotemporal dynamic changes of EEG corresponding to the memory encoding stage, storage stage, and retrieval stage through the display device 300.

[0106] At the same time, the display device 300 shows the memory encoding ability score, memory storage ability score, memory retrieval ability score, and memory ability level results, thereby generating a corresponding memory ability assessment report.

[0107] Please see Figure 8 This is an example diagram of a memory ability assessment report according to an embodiment of the present invention.

[0108] Through the above steps, a complete assessment process is realized, from EEG signal acquisition, signal preprocessing, feature extraction, brain-behavior association analysis, memory ability assessment to result visualization, thereby completing the automated assessment of children's and adolescents' memory abilities.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A brainwave-based memory ability assessment system for children and adolescents, characterized in that, include: EEG acquisition equipment is used to collect EEG signals from subjects during the performance of working memory tasks, and simultaneously collect behavioral data; The processing terminal is communicatively connected to the EEG acquisition device and is used to receive the EEG signals and behavioral data, and to perform memory ability assessment processing. A display device, connected to the processing terminal, is used to display the evaluation results; The processing terminal includes: The preprocessing module is used to filter, segment, correct baselines, remove artifacts, and configure reference electrodes on the EEG signal to obtain a clean EEG signal. The feature extraction module is used to extract the spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features corresponding to the memory encoding stage, memory storage stage, and memory retrieval stage from the clean EEG signal, and to construct full-dimensional spatiotemporal core features. The brain-behavior mapping module is used to correlate and fit the full-dimensional spatiotemporal core features with behavioral data to establish a quantitative correlation model between EEG features and memory encoding ability, memory storage ability and memory retrieval ability. The assessment module is used to assess the memory ability of the target subjects based on the quantitative correlation model, and obtain memory encoding ability score, memory storage ability score, memory retrieval ability score and comprehensive memory ability score. The spatiotemporal display module is used to generate dynamic topographic maps, dynamic brain network maps, and three-dimensional source activation maps based on the full-dimensional spatiotemporal core features, and to visualize them in conjunction with the memory ability assessment results.

2. The EEG-based memory ability assessment system for children and adolescents according to claim 1, characterized in that, The feature extraction module extracts full-dimensional spatiotemporal core features according to the memory encoding stage, memory storage stage, and memory extraction stage respectively. The full-dimensional spatiotemporal core features corresponding to the memory encoding stage include the spatiotemporal distribution features of scalp EEG energy; The full-dimensional spatiotemporal core features corresponding to the memory storage stage include brain network functional connectivity features; The full-dimensional spatiotemporal core features corresponding to the memory retrieval stage include cortical source activation features.

3. The EEG-based memory ability assessment system for children and adolescents according to claim 1, characterized in that, The preprocessing module obtains the clean EEG signal through the following steps: filtering the original EEG signal using a 0.5Hz to 50Hz bandpass filter; dividing the continuous EEG signal into EEG data segments corresponding to preset time windows according to the stimulus event markers; performing baseline correction using a preset baseline time window; and identifying and removing artifacts such as electrooculogram (EOG), electrocardiogram (ECG), and electromyogram (EMG). Rereference processing is performed on the EEG signals.

4. The EEG-based memory ability assessment system for children and adolescents according to claim 1, characterized in that, The spatiotemporal distribution characteristics of scalp EEG energy include theta band power characteristics, alpha band power characteristics, and the amplitude and latency characteristics of event-related potentials; The brain network functional connectivity features are obtained through the functional connectivity matrix between brain electrical channels, and the clustering coefficient and global efficiency are calculated based on the functional connectivity matrix. The cortical source activation features are obtained through brain power imaging and include the cortical source activation time series corresponding to the target brain region.

5. The EEG-based memory ability assessment system for children and adolescents according to claim 1, characterized in that, The brain-behavior mapping module uses the ridge regression algorithm to establish the quantitative correlation model; The full-dimensional spatiotemporal core features are constructed into an EEG feature matrix; The behavioral data is then constructed into a behavioral target vector; By introducing an L2 regularization constraint term into the ridge regression objective function, the mapping relationship between EEG features and behavioral indicators is obtained, thereby reducing the impact of the correlation between the full-dimensional spatiotemporal core features on the model training results.

6. A method for assessing memory ability in children and adolescents based on electroencephalography (EEG), characterized in that, include: Collect electroencephalogram (EEG) signals and behavioral data from subjects during the performance of working memory tasks; The EEG signals are preprocessed to obtain clean EEG signals; Extract the spatiotemporal distribution features of scalp EEG energy, brain network functional connectivity features, and cortical source activation features corresponding to the memory encoding stage, memory storage stage, and memory retrieval stage from the clean EEG signals, and construct full-dimensional spatiotemporal core features. A quantitative correlation model between EEG features and behavioral indicators is established based on the full-dimensional spatiotemporal core features and the behavioral data. Based on the quantitative correlation model, the memory ability of the target subjects is assessed to obtain memory encoding ability score, memory storage ability score, memory retrieval ability score and comprehensive memory ability score; A visual display result is generated based on the full-dimensional spatiotemporal core features and the memory ability assessment results.

7. The method for assessing children's and adolescents' memory ability based on electroencephalography (EEG) according to claim 6, characterized in that, The extracted cortical source activation features include: Establish a brain power source imaging model between scalp EEG observation signals and cortical source signals; The observation residuals in the brain power imaging model are modeled as a Student t-distribution to characterize non-Gaussian noise and outliers in the EEG signal; The student t-distribution is approximated by a Gaussian mixture distribution, and variational parameters are introduced to construct the observation noise covariance matrix; Based on the observed noise covariance matrix and the prior distribution of the cortical source signal, the posterior distribution of the cortical source signal is obtained by Bayesian inference. The cortical source activation time series is determined based on the posterior distribution, and the cortical source activation time series is used as the cortical source activation feature.

8. The method for assessing children's and adolescents' memory ability based on electroencephalography (EEG) according to claim 7, characterized in that, The approximate representation of the student t-distribution using a Gaussian mixture distribution includes: Introduce variational parameters corresponding to each observation channel; The student t-distribution can be expressed as a multiplicative distribution of the Gaussian and inverse gamma distributions using the variational parameters. A diagonal observation noise covariance matrix is ​​constructed based on the variational parameters to characterize the noise level corresponding to different EEG channels, wherein the observation noise covariance matrix is ​​a diagonal matrix composed of the variational parameters of each observation channel.

9. The method for assessing children's and adolescents' memory ability based on electroencephalography (EEG) according to claim 7, characterized in that, The method of obtaining the posterior distribution of cortical source signals using Bayesian inference includes: Construct a probabilistic model with a zero-mean Gaussian distribution as the prior distribution of cortical source signals; The posterior mean and posterior covariance matrix of the cortical source signal are calculated based on the observed EEG signal, lead field matrix, observed noise covariance matrix, and the prior distribution. The model parameters are iteratively updated by maximizing the log-likelihood function of the complete data until the preset convergence condition is met. The converged posterior mean is used as the robust estimation result of the cortical source signal, and the corresponding cortical source activation time series is generated. The cortical source activation time series, together with the spatiotemporal distribution characteristics of scalp EEG energy and brain network functional connectivity characteristics, constitute the full-dimensional spatiotemporal core features, and are used to establish a quantitative correlation model between EEG features and behavioral indicators.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the EEG-based method for assessing the memory abilities of children and adolescents as described in any one of claims 6 to 9.

Images